The abundances of short-lived radionuclides in the early solar system (ESS) are reviewed, as well as the methodology used in determining them. These results are compared with the inventory estimated for a uniform galactic production model. It is shown that, to within a factor of two, the observed abundances of 238 U, 235 U, 232 Th, 244 Pu, 182 Hf, 146 Sm, and 53 Mn are roughly compatible with long-term galactic nucleosynthesis. 129 I is an exception, with an ESS inventory much lower than expected from uniform production. The isotopes 107 Pd, 60 Fe, 41 Ca, 36 Cl, 26 Al, and 10 Be require late addition to the protosolar nebula. 10 Be is the product of energetic particle irradiation of the solar system as most probably is 36 Cl. Both of these nuclei appear to be present when 26 Al is absent. A late injection by a supernova (SN) cannot be responsible for most of the short-lived nuclei without excessively producing 53 Mn; it can however be the source of 53 Mn itself and possibly of 60 Fe. If a late SN injection is responsible for these two nuclei, then there remains the problem of the origin of 107 Pd and several other isotopes. Emphasis is given to an AGB star as a source of many of the nuclei, including 60 Fe; this possibility is explored with a new generation of stellar models. It is shown that if the dilution factor (i.e. the ratio of the contaminating mass to the solar parental cloud mass) is f 0 ∼ 4×10 −3 , a reasonable representation for many nuclei is obtained; this requires that ( 60 Fe/ 56 Fe) ESS ∼ 10 −7 to 2×10 −6 . The nuclei produced by an AGB source do not include 53 Mn, 10 Be or 36 Cl if it is very abundant. The role of irradiation is discussed with regard to 26 Al, 36 Cl and 41 Ca, and the estimates of bulk solar abundances of these isotopes are commented on. The conflict between various scenarios is emphasized as well as the current absence of an astrophysically plausible global interpretation for all the existing data. Examination of abundances for the actinides indicates that a quiescent interval of ∼ 10 8 years is required for actinide group production. This is needed in order to explain the data on 244 Pu and the new bounds on 247 Cm. Because this quiescent interval is not compatible with the 182 Hf data, a separate type of r-process event is needed for at least the actinides, distinct from the two types that have previously been identified. The apparent coincidence of the 129 I and trans-actinide time scales suggests that the last heavy r contribution was from an r-process that produced very heavy nuclei but without fission recycling so that the yields at Ba and below (including I) were governed by fission.